The UT Internet Poll (again)
Mon March 23, 2009 4:21 pm

This is my response to the post, below, of the government professors who operate the UT Poll. I have no desire to get into a verbal wrestling match with Professors Henson and Shaw. I have spoken to both of them, in person and by telephone, and I respect their work and what they are trying to accomplish. They are entitled to run a poll the way the want to run it, and I am entitled to give my opinion of its worth.

1. In my opinion, Internet polling is not a proven methodology. In particular, Zogby's Internet polls are useless. There are too many things about Internet polling that are eccentric. Would you put more faith in a poll of registered voters if you determined a person's status from an Internet interview or from a voter registration list? Internet polling does not use interviews. It locates participants.

2. I did not write about the part of the poll that involved the U.S. Senate race. The poll presented participants with brief biographies of selected candidates. How in the world could the pollsters control for bias? If this is not a push poll, it is certainly a cousin.

Here is the bio for Roger Williams:

"Roger Williams (Republican) is a former Texas Secretary of State and served as chair of the Texas Republican Party’s 2008 campaign committee. He is a former college baseball star from Texas Christian University who embarked on a career in business and public affairs after playing and coaching baseball." (No mention that he is a car dealer.)

3. According to the web site for the poll, these three questions were asked of the participants:

Q1. Are you registered to vote in the state of Texas?
1. Yes 90%
2. No 10%
3. Don’t know 0%

--My question to readers: Do you think that a normal sample of 800 Texans would include 90% registered voters? I don't.

Q2. On a scale of 0 to 10, where 0 is extremely uninterested, 10 is extremely interested, and 5 is exactly in the middle, how interested, generally speaking, would you say you are politics and public affairs? Enter a number between 0 and 10.

7.3 mean

--My question to readers: Do you think that a normal sample of 800 Texans would find that the typical respondent follows politics and public affairs at an interest level roughly halfway between average and "extremely closely?" I doubt it.

Q3. On a scale of 0 to 10, where 0 is not at all, 10 is extremely closely, and 5 is exactly in the middle, how closely would you say you will be following the Texas legislative session this year? Enter a number between 0 and 10.

5.2 mean.

--My question to readers: Do you think that a normal sample of 800 Texans would show that more than half of them are following the Texas legislative session? I don't. Walk 200 feet from the Capitol in any direction and ask anyone on the street what is going on. Unless lobbyists are out getting some exercise, nobody will have clue.

This was my basic point. The participants in the poll have more than the normal interest in politics. And that has to skew the results.

More Texas Monthly

Loading, please wait...

Most Read

  • Viewed
  • Past:
  • 1 week